Date post: | 20-Feb-2018 |
Category: |
Documents |
Upload: | tumballisto |
View: | 217 times |
Download: | 0 times |
of 10
7/24/2019 jeas_0315_1653
1/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1701
MULTIPLE CRACK DETECTION IN BEAMS FROM THE DIFFERENCES
IN CURVATURE MODE SHAPES
K. Ravi Prakash Babu1, B. Raghu Kumar2, K. L.Narayana2 and K. Mallikarjuna Rao11Deptartment of Mechanical Engineering, JNTUK Kakinada, India
2K L E F, K.L. University, Guntur, India
E-Mail: [email protected]
ABSTRACT
The presence of crack in a structure tends to modify its modal parameters (natural frequencies and mode shapes).The fact can be used inversely to predict the crack parameters (crack depth and its location) from measurement of the
changes in the modal parameters, once a functional relationship between them has been established. The machine
components like turbine blade can be treated as a cantilever beam and a shaft as a simply supported beam. Vibrationanalysis of cantilever beam and simply supported beam can be extended successfully to develop online crack detection
methodology in turbine blades and shafts. In the present work, finite element analysis of a cantilever and simply supported
beams for flexural vibrations has been considered by including two transverse open U-notches. The modal analysis has
been carried out on cantilever and simply supported beams with two U-notches and observed the influence of one U-notch
on the other for natural frequencies and mode shapes. This has been done by carrying out parametric studies using ANSYSsoftware to evaluate the natural frequencies and their corresponding mode shapes for different notch parameters (depths
and locations) of the cantilever and simply supported beams FEM model. Later, by using a central difference
approximation, curvature mode shapes were then calculated from the displacement mode shapes. The location and depthcorresponding to any peak on this curve becomes a possible notch location and depth. The identification procedure
presented in this study is believed to provide a useful tool for detection of medium size crack in a cantilever and simply
supported beam applications.
KEYWORDS:crack detection, vibration, FEM, displacement mode shapes, curvature mode shapes.
INTRODUCTION
Any localized crack in a structure reduces thestiffness in that area. These features are related to variation
in the dynamic properties, such as, decreases in natural
frequencies and variation of the modes of vibration of thestructure. One or more of above characteristics can be
used to detect and locate cracks. This property may beused to detect existence of crack or faults together with
location and its severity in a structural member. Rizos [1]
measured the amplitude at two points and proposed analgorithm to identify the location of crack. Pandey [2]
suggested a parameter, namely curvature of the deflected
shape of beam instead of change in frequencies to identify
the location of crack. Ostachowicz [3] proposed aprocedure for identification of a crack based on the
measurement of the deflection shape of the beam.
Ratcliffe [4] also developed a technique for identifying thelocation of structural damage in a beam using a 1D FEA.
A finite difference approximation called Laplacesdifferential operator was applied to the mode shapes to
identify the location of the damage. Wahab [5]investigated the application of the change in modal
curvatures to detect damage in a prestressed concrete
bridge. Lakshminarayana [6] carried out analytical work tostudy the effect of crack at different location and depth on
mode shape behaviour. Chandra Kishen [7] developed a
technique for damage detection using static test data.
Nahvi [8] established analytical as well as experimentalapproach to the crack detection in cantilever beams. Ravi
Prakash Babu [9] used differences in curvature mode
shapes to detect a crack in beams.
This paper deals with the technique and its
application of mode shapes to a cantilever and a simplysupported beam. The paper of Pandey et al. [2] shows a
quite interesting phenomenon: that is, the modal
curvatures are highly sensitive to crack and can be used tolocalize it. They used simulated data for a cantilever and a
simply supported beam model with single damage todemonstrate the applicability of the method. The cracked
beam was modeled by reducing the E-modulus of a certain
element. By plotting the difference in modal curvaturebetween the intact and the cracked beam, a peak appears at
the cracked element indicating the presence of a fault.
They used a central difference approximation to derive the
curvature mode shapes from the displacement modeshapes. An important remark could be observed from the
results of Pandey et al. [2]: that is, the difference in modal
curvature between the intact and the damaged beamshowed not only a high peak at the fault position but also
some small peaks at different undamaged locations for thehigher modes. To avoid this, a U-notch was modelled at
the location of faults instead of reducing the E-modulus ofa certain element. Also, analysis was carried out for beams
with two cracks instead of one.
So, this paper is done to study about the changesin mode shapes because of the presence of U-notches in
the cantilever and simply supported beams and is
concerned with investigation of the accuracy when using
the central difference approximation to compute the modalcurvature and determine the location of the U-notches and
to find out the reason of the presence of the misleading
small peaks. The application of this technique to
7/24/2019 jeas_0315_1653
2/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1702
constructions in which more than one fault positions existis investigated using a continuous beam with simulated
data. The results of this scenario will be analyzed in this
paper and U-notches will be detected and localized by
using the measured change in modal curvatures.So, as a summary, in the present work, a methodology for
predicting crack parameters (crack depth and its location)
in a cantilever and simply supported beam from changes incurvature mode shapes has been developed. Parametric
studies have been carried out using ANSYS Software to
evaluate mode shapes for different crack parameters
(depth and its location). Curvature mode shapes were then
calculated from the displacement mode shapes to identify
crack location and its severity in the cantilever and simplysupported beam.
PROBLEM FORMULATION
Figures-1(a) & 1(b) shows a cantilever and asimply supported beam of rectangular cross section, made
of mild steel with two U- notches. To find out mode
shapes associated with each natural frequency, FE analysishas been carried out using ANSYS Software for un-
notched and U-notched beam.
Figure-1(a). Cantilever beam with two U-notch cracks.
Figure-1(b). Simply supported beam with two U-notch cracks.
The mode shapes of the multiple U-notched
cantilever and simply supported beams are obtained for U-notches located at normalized distances (c/l & d/l) from
the fixed end of a cantilever beam and from left end
support of a simply supported beam with a normalizeddepth (a/h).
Figure-2. Discretised model of beam with two U-notches.
Figure-2 shows the discretised model (zoomed
near the position of U-notches) of a beam with two U-notches. Parametric studies have been carried out for thin
beam having length (l) = 260 mm, width (w) = 25 mm and
thickness (h) = 4.4 mm. The breadth (b) of each U-notchhas been kept as 0.32 mm. The U-notch locations from the
fixed end (c & d) of the cantilever beam have been taken
in different combinations near the fixed end, free end andmiddle of the beam. Similarly, U-notch locations from theleft end support (c & d) of the simply supported beam
have been taken in different combinations near the
supports and middle of the beam. The intensity of U-notch(a/h) was varied by increasing its depth over the range of
0.25 to 0.75 in the steps of 0.25. This represented the case
of a varying degree of crack at particular location. For
each model of the two U-notch locations, the first three
natural frequencies and corresponding mode shapes werecalculated using ANSYS software.
7/24/2019 jeas_0315_1653
3/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1703
Governing equation
Determination of modal parameters (natural
frequency & mode shape) in a beam is an Eigen value
problem. ANSYS Software is used for theoretical modalanalysis of the beam and the governing equation for
general Eigen value problem is:
0 XKXM (1)Disposing the brackets without ambiguity equation (1) is
rewritten as follows:
0KXXM (2)
Pre-multiplying both sides of equation (2) by1
M :
0111 MKXMXMM Now, XIXMM
1and
AXKXM 1
(say)
0AXXI (3)
Where, KMA1
= system matrix.
Assuming harmonic motion ;XX where, 2
equation (3) becomes
0 XIA (4)The characteristic of motion is then:
0 IA (5)
The n-rootsi
, where i=1, 2, 3n of the characteristic
equation (5) are called Eigen values.
The natural frequencies are found as:
ii , i = 1,2,3n. (6)
Substitution of the Eigen values (i
s) in (4) gives the
mode shapesi
X corresponding toi
. These are Eigen
vectors.
RESULTS AND DISCUSSION
In this analysis, it is assumed that crack is of U-
notched shape. The depth (a) and locations (c & d) ofthese notches are normalized to the height and length of
the cantilever and simply supported beams respectively.
The first three mode shapes for the beam were calculatedusing ANSYS software and were shown below for
different crack depths and crack location ratios.
Curvature finite difference approximationLocalized changes in stiffness result in a mode
shape that has a localized change in slope, therefore, this
feature will be studied as a possible parameter for crack
detection purposes. For a beam in bending the curvature(k) can be approximated by the second derivate of the
deflection:
(7)
In addition, numerical mode shape data is discretein space, thus the change in slope at each node can be
estimated using finite difference approximations. In thiswork, the central difference equation was used to
approximate the second derivate of the displacements ualong the x - direction at node i:
(8)
The term is the element length.
In this process meshing and node numbering is very
important. Equation (8) require the knowledge of thedisplacements at node i, node i -1 and node i+1 in order to
evaluate the curvature at node i. Thus, the value of the
curvature of the mode shapes could be calculated starting
from node 2 through node 261 in case of this beam. After
obtaining the curvature mode shapes the absolute
difference between the uncracked and cracked state isdetermined to improve crack detection.
(9)
As a result of this analysis, a set of curvature
vectors for different crack localizations are obtained.
Uncracked case
By using the same finite element model shown inFigure-2, linear mode shapes was performed in ANSYS.
The numerical results were exported to MATLAB to be
processed.The associated mode shapes were sketched
evaluating the displacements in y direction of the 261
equidistant nodes located at the bottom line of the beam.In order to unify the results from the different cases, mode
shapes were normalized by setting the largest grid point
displacement equal to 1. It can be noticed from figures thatall the mode shapes smoothed functions, what indicate the
absence of cracks. Cracked beam mode shapes will be
used to compare further results. Since changes in the
curvature are local in nature, they can be used to detect
and locate cracks in the beam.
Simple cracked case
The different combinations of crack locationscenarios are selected for studying the effect of localized
cracks in the beam. Although the reduction in natural
frequencies is related to the existence of crack and itsseverity, this feature cannot provide any usefulinformation about the location of the crack. Thus,
curvature mode shapes were calculated and compared with
the uncracked case. It can be seen that the maximumdifference value for each mode shape occurs in the crack
locations. In other areas of the beam this characteristic was
much smaller. Although the third mode shapes was the
most sensitive to the failure it is important that any of the
three curvature mode shapes peak at the cracked locations.
7/24/2019 jeas_0315_1653
4/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1704
Cantilever beam
Figure-3. Mode shapes for crack at c/l = 0.15 of depth a/h = 0.25 and crack at d/l = 0.8 of depth a/h = 0.75.
Figure-4. Difference curvature for crack at c/l = 0.15 of depth a/h = 0.25 and crack at d/l = 0.8 of depth a/h = 0.75.
Figure-5. Mode shapes for crack at c/l = 0.15 of depth a/h = 0.5 and crack at d/l = 0.8 of depth a/h = 0.5.
7/24/2019 jeas_0315_1653
5/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1705
Figure-6. Difference curvature for crack at c/l = 0.15 of depth a/h = 0.5 and crack at d/l = 0.8 of depth a/h = 0.5.
Figure-7. Mode shapes for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.65 of depth a/h = 0.5.
Figure-8. Difference curvature for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.65 of depth a/h = 0.5.
7/24/2019 jeas_0315_1653
6/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1706
Figure-9. Mode shapes for crack at c/l = 0.25 of depth a/h = 0.75 and crack at d/l = 0.65 of depth a/h = 0.5.
Figure-10. Difference curvature for crack at c/l = 0.25 of depth a/h = 0.75 and crack at d/l = 0.65 of depth a/h = 0.5.
To examine the curvature mode shape techniquefor cantilever beam having several crack locations, the
span of the beam is discretized by 260 elements. Two
crack locations are assumed at a time. Firstly, two cracks
at c/l = 0.15 of depth a/h = 0.25 and at c/l = 0.8 of depth
a/h = 0.75 were considered. The first three displacementmode shapes are shown in Figure 3. The difference inmodal curvature between the uncracked and the cracked
beam is plotted in Figure 4 for the first three modes. For
mode 1 in Figure 4, it can be observed that the peak at c/l= 0.15 is very small comparing to that at d/l = 0.8. And
also, for mode 1 the modal curvature at c/l = 0.15 is less
and has very small values at the nodes next to it. Incontrast, at d/l = 0.8 high modal curvature takes place.
This indicates the severity of the crack depth ratio a/h =0.75 at d/l = 0.8 compared to crack depth ratio a/h = 0.25
at c/l = 0.15. And also, first mode shape modal curvature
is more sensitive near the fixed end compared to second
and third mode shape modal curvatures. Again second and
third mode shape modal curvatures are more sensitive nearthe free end compared to first mode shape modalcurvature. The same can be observed for some other crack
scenarios shown in Figures-5 & 6, 7 & 8, 9 & 10. So,
depending on the absolute ratio between the modalcurvature values for a particular mode at two different
locations, one peak can dominate the other. Therefore, one
can conclude that in case of several crack locations in astructure, all modes should be carefully examined.
7/24/2019 jeas_0315_1653
7/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1707
Simply supported beam
Figure-11. Mode shapes for crack at c/l = 0.15 of depth a/h = 0.25 and crack at d/l = 0.8 of depth a/h = 0.5.
Figure-12. Difference curvature for crack at c/l = 0.15 of depth a/h = 0.25 and crack at d/l = 0.8 of depth a/h = 0.5.
Figure-13. Mode shapes for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.45 of depth a/h = 0.75.
7/24/2019 jeas_0315_1653
8/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1708
Figure-14. Difference curvature for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.45 of depth a/h = 0.75.
Figure-15. Mode shapes for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.65 of depth a/h = 0.5.
Figure-16. Difference curvature for crack at c/l = 0.25 of depth a/h = 0.5 and crack at d/l = 0.65 of depth a/h = 0.5.
7/24/2019 jeas_0315_1653
9/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1709
Figure-17. Mode shapes for crack at c/l = 0.65 of depth a/h = 0.5 and crack at d/l = 0.85 of depth a/h = 0.5.
Figure-18. Difference curvature for crack at c/l = 0.65 of depth a/h = 0.5 and crack at d/l = 0.85 of depth a/h = 0.5.
To examine the curvature mode shape technique
for simply supported beam having several crack locations,
the span of the beam is also discretized by 260 elements.
Two crack locations are assumed at a time. Firstly, two
cracks at c/l = 0.15 of depth a/h = 0.25 and at c/l = 0.8 ofdepth a/h = 0.5 were considered. The first three
displacement mode shapes are shown in Figure 11. Thedifference in modal curvature between the uncracked and
the cracked beam is plotted in Figure 12 for the first three
modes. For all the modes in Figure 12, it can be observed
that the peak at c/l = 0.15 is very small comparing to thatat d/l = 0.8. In contrast, at d/l = 0.8 high modal curvature
takes place. This indicates the severity of the crack depth
ratio a/h = 0.5 at d/l = 0.8 compared to crack depth ratioa/h = 0.25 at c/l = 0.15. And also, third mode shape modal
curvature is more sensitive near the supports compared to
first and second mode shape modal curvatures. The same
can be observed in Figure 18 at d/l = 0.85. Again in Figure
14 third mode shape modal curvature is more sensitive
near the middle portion of the beam at d/l = 0.45 compared
to other mode shape modal curvature and the second mode
modal curvature is more sensitive near the one-fourth of
the length of the beam at c/l = 0.25. The same can be
observed for some other crack scenarios shown in Figure16 at d/l = 0.65 and in Figure 20 at c/l = 0.65 for second
mode shape. So, depending on the absolute ratio betweenthe modal curvature values for a particular mode at two
different locations, one peak can dominate the other.
Therefore, one can conclude that in case of several crack
locations in a structure, all modes should be carefullyexamined.
And also, the crack is assumed to affect stiffness
of the cantilever beam. The stiffness matrix of the crackedelement in the FEM model of the beam will replace the
stiffness matrix of the same element prior to damaging to
result in the global stiffness matrix. Thus frequencies and
mode shapes are obtained by solving the eigen value
problem [K] 2 [M] = 0. So it can be seen in Figure 3, 9,
7/24/2019 jeas_0315_1653
10/10
VOL. 10, NO. 4, MARCH 2015 ISSN 1819-6608
ARPN Journal of Engineering and Applied Sciences
2006-2015 Asian Research Publishing Network (ARPN). All rights reserved.
www.arpnjournals.com
1710
13 very clearly the changes in slopes and deviations inmode shape at crack location for crack depth ratio 0.75.
CONCLUSIONS
A method for identifying multiple crackparameters (crack depth and its location) in beams using
modal parameters has been attempted in the present paper.
Parametric studies have been carried out using ANSYSSoftware to evaluate modal parameters (natural
frequencies and mode shapes) for different crack
parameters. A theoretical study using simulated data for a
cantilever and simply supported beams has been
conducted. When more than one fault exists in thestructure, it is not possible to locate crack in all positions
from the results of only one mode. All modes should be
carefully examined in order to locate all existing faults.Also, due to the irregularities in the measured mode
shapes, a curve fitting can be applied by calculating the
curvature mode shapes using the central differenceapproximation. The curvature mode shapes technique forcrack localization in structures is investigated in this
paper. The results confirm that the application of the
curvature mode shape method to detect cracks inengineering structures seems to be promising. Techniques
for improving the quality of the measured mode shapes are
highly recommended. The identification procedure
presented in this paper is believed to provide a useful tool
for detection of medium size cracks in beams.
ACKNOWLEDGEMENTS
The author is grateful to the President, K L E FUniversity, for the kind permission to publish the paper. I
am also grateful to all the staff members of Mechanicaldepartment who have helped in carrying out the work.
REFERENCES
[1] P. F. Rizos., N Aspragathos. and A. D. Dimarogonas.
1990. Identification of crack location and magnitude
in a cantilever beam from the vibration modes,
Journal of Sound and Vibration. 138: 381-388.
[2] A. K. Pandey., M. Biswas and M. M. Samman. 1991.
Damage detection from change in curvature mode
shapes, Journal of Sound and Vibration. 145: 321-332.
[3]
W. M.Ostachowicz. and M. Krawczuk. 1991.Analysis of the effect of cracks on natural frequencies
of a cantilever beam, Journal of Sound and Vibration.
150: 191-201.
[4] C. P. Ratcliffe. 1997. Damage detection using a
modified Laplacian operator on mode shape data,
Journal of Sound and Vibration. 204(3): 505-517.
[5] M. Wahab and G Roeck. 1999. Damage detection inbridges using modal curvatures: application to a real
damage scenario. Journal of Sound and Vibration.226(2):217-235.
[6] K. Lakshminarayana. and C. Jebaraj. 1991. Sensitivity
analysis of local/global modal parameters foridentification of a crack in a beam. Journal of Sound
and Vibration. 228: 977-994.
[7] J. M. Chandra Kishen and Trisha Sain. 2004. Damage
detection using static test data, Journal of Structural
Engineering, India. 31(1): 15-21.
[8] H. Nahvi. and M. Jabbari. 2005. Crack detection inbeams using experimental modal data and finite
element model. International Journal of Mechanical
Sciences. 47: 1477-1497.
[9] K. Ravi Prakash Babu and G. Durga Prasad. 2012.
Crack detection in beams from the differences incurvature mode shapes. Journal of Structural
Engineering, India. 39(2): 237-242.